Classifying space for SO(n) explained

In mathematics, the classifying space

\operatorname{BSO}(n)

for the special orthogonal group

\operatorname{SO}(n)

is the base space of the universal

\operatorname{SO}(n)

principal bundle

\operatorname{ESO}(n)\operatorname{BSO}(n)

. This means that

\operatorname{SO}(n)

principal bundles over a CW complex up to isomorphism are in bijection with homotopy classes of its continuous maps into

\operatorname{BSO}(n)

. The isomorphism is given by pullback.

Definition

There is a canonical inclusion of real oriented Grassmannians given by

k+1
\widetilde\operatorname{Gr}
n(R

), V\mapstoV x \{0\}

. Its colimit is:[1]
infty) :=\lim
\operatorname{BSO}(n) :=\operatorname{Gr}
k → infty
k).
\widetilde\operatorname{Gr}
n(R

Since real oriented Grassmannians can be expressed as a homogeneous space by:

k) =\operatorname{SO}(n+k)/(\operatorname{SO}(n) x \operatorname{SO}(k))
\widetilde\operatorname{Gr}
n(R

the group structure carries over to

\operatorname{BSO}(n)

.

Simplest classifying spaces

\operatorname{SO}(1) \cong1

is the trivial group,

\operatorname{BSO}(1) \cong\{*\}

is the trivial topological space.

\operatorname{SO}(2) \cong\operatorname{U}(1)

, one has

\operatorname{BSO}(2) \cong\operatorname{BU}(1) \congCPinfty

.

Classification of principal bundles

X

the set of

\operatorname{SO}(n)

principal bundles on it up to isomorphism is denoted

\operatorname{Prin}\operatorname{SO(n)}(X)

. If

X

is a CW complex, then the map:[2]

[X,\operatorname{BSO}(n)]\operatorname{Prin}\operatorname{SO(n)}(X), [f]\mapstof*\operatorname{ESO}(n)

is bijective.

Cohomology ring

The cohomology ring of

\operatorname{BSO}(n)

with coefficients in the field

Z2

of two elements is generated by the Stiefel–Whitney classes:[3] [4]
*(\operatorname{BSO}(n);Z
H
2) =Z

2[w2,\ldots,wn].

\operatorname{char}=2

.

The cohomology ring of

\operatorname{BSO}(n)

with coefficients in the field

Q

of rational numbers is generated by Pontrjagin classes and Euler class:
*(\operatorname{BSO}(2n);Q) \congQ[p
H
1,\ldots,p

n,e]/(p

2),
n-e
*(\operatorname{BSO}(2n+1);Q) \congQ[p
H
1,\ldots,p

n].

The results holds more generally for every ring with characteristic

\operatorname{char}2

.

Infinite classifying space

The canonical inclusions

\operatorname{SO}(n)\hookrightarrow\operatorname{SO}(n+1)

induce canonical inclusions

\operatorname{BSO}(n)\hookrightarrow\operatorname{BSO}(n+1)

on their respective classifying spaces. Their respective colimits are denoted as:

\operatorname{SO} :=\limn → infty\operatorname{SO}(n);

\operatorname{BSO} :=\limn → infty\operatorname{BSO}(n).

\operatorname{BSO}

is indeed the classifying space of

\operatorname{SO}

.

See also

Literature

External links

References

  1. Milnor & Stasheff 74, section 12.2 The Oriented Universal Bundle on page 151
  2. Web site: 2024-03-14 . en . universal principal bundle . nLab.
  3. Milnor & Stasheff, Theorem 12.4.
  4. Hatcher 02, Example 4D.6.